By Douglas L. Kriner
This page provides the interested reader with additional details about how we conducted the study discussed in the Nationhood Lab article “Regional differences in perceptions to the threats to U.S. democracy,” as well as additional analyses that show estimates of statistical uncertainty or that report results from statistical models that assess differences across geographies controlling for differences in partisanship and demographics.
The survey was conducted in partnership between the Institute of Politics and Global Affairs in the Jeb E. Brooks School of Public Policy at Cornell University, GQR Insights and Action and Verasight. Verasight employed a mixture of recruitment strategies including dynamic online sampling and random address-based sampling to produce our three survey samples: a general population adult national sample (n = 1,316); a representative sample of adults in Michigan’s 8th congressional district (n = 550); a representative sample of adults in Texas 15th congressional district (n = 573). More information about Verasight’s sampling methodology is available here. The survey was fielded from February 17 – April 1, 2022. GQR calculated sampling weights for all three survey samples. However, because we do not have weights specifically constructed for each American nation, all percentages and statistical analyses reported in this post do not use survey weights. The number of respondents in the national sample by American nation are presented in the table below. While all respondents in the national sample are used to calculate the national average for each question, we omit nations with fewer than 30 respondents from any comparative analyses.
| Nation | Frequency | Percent |
| Deep South | 202 | 15.35 |
| El Norte | 143 | 10.87 |
| Far West | 90 | 6.84 |
| Greater Appalachia | 273 | 20.74 |
| Left Coast | 53 | 4.03 |
| Midlands | 128 | 9.73 |
| New Netherland | 58 | 4.41 |
| Tidewater | 33 | 2.51 |
| Yankeedom | 287 | 21.81 |
| Other | 49 | 3.72 |
Below, we provide additional details and analyses to provide greater information on specific points made in the post.
End Note #1 — Re: Percentages believing January 6th was a major threat to democracy by nation
The figure below presents the percentages believing January 6 a major threat to democracy by region with 90% confidence intervals about each percentage. The horizontal line at 51% is the corresponding percentage among the national sample as a whole. We note that the 90% confidence intervals for Midlands, New Netherland, and Tidewater all include the estimate for the national sample as a whole. However, combining data from these three regions shows that the percentage believing January 6 a major threat (58%) is significantly higher than the national average (51%). Throughout, we define statistical significant as p < .05, one-tailed test.
End Note #2 — RE: Percentages believing efforts to change laws to make it hard to vote is a major threat to democracy by nation
The figure below shows the percentages believing changing laws to make it harder to vote is a major threat to democracy by nation with 90% confidence intervals around each percentage. The vertical line at 51% shows the national average. The percentages believing this a major threat in Deep South, El Norte, and Greater Appalachia are all lower than the national average; however, the 90% confidence intervals for each include the national average. Combining data from these three nations, we find that the percentage believing this a major threat in these nations (47%) is significantly lower than the national average (51%). Similarly, the percentage believing this a major threat in Left Coast, Tidewater, and Midlands is all higher than the national average. The confidence intervals for both Left Coast and Tidewater include the national average; however, combining these three nations shows that the percentage believing this a major threat in these nations (61%) is significantly higher than the national average (51%).
End Note #3 — RE: Percentages believing social media censorship is a major threat to democracy by nation
The figure below shows the percentages believing changing laws to make it harder to vote is a major threat to democracy by nation with 90% confidence intervals around each percentage. The vertical line at 43% shows the national average.
Note: I-bars show 90% confidence intervals around each percentage.
End Note #4 — RE: Percentages believing voter fraud is a major threat to democracy by nation
The figure below shows the percentages believing changing laws to make it harder to vote is a major threat to democracy by nation with 90% confidence intervals around each percentage. The vertical line at 37% shows the national average.
End Note #5 — RE: Percentages believing efforts to “cancel” history are a major threat to democracy by nation
The figure below shows the percentages believing changing laws to make it harder to vote is a major threat to democracy by nation with 90% confidence intervals around each percentage. The vertical line at 50% shows the national average. In the post, we mention that a majority of respondents in Deep South, Greater Appalachia, Midlands, and Yankeedom identified this as a major threat, while only minorities did so in El Norte, Far West, Left Coast, New Netherland, and Tidewater. On average, 54% of respondents believed efforts to cancel history were a major threat to democracy in the first set of four nations vs. 41% in the latter five nations; this difference is highly statistically significant.
Note: I-bars show 90% confidence intervals around each percentage.
End Note #6 — RE: Differences in percentages believing each a major threat and supporting voter ID laws between swing CDs and the national sample.
Interestingly, residents of TX-15 were more concerned than the national sample on average about a range of threats including some concerning primarily to the political left (e.g. January 6, efforts to disenfranchise voters of color) and those of greater concern to the political right (e.g. canceling speech and social media censorship). TX-15 residents were also significantly more supportive of voter ID laws than the national sample (69% vs. 59%). Residents of MI-8 were more likely to view all threats as a major threat to democracy except for canceling history, social media censorship, and voter fraud – all three of which are primarily of concern to ideological conservatives.
End Note #7 — RE: Differences in democracy threat perceptions and support for voter ID laws between swing CDs and the rest of their nation
In the post, we reference raw differences in the percentages believing different threats pose a major threat for democracy across the two swing congressional districts and their respective nations. However, to examine the extent to which these differences may simply be the result of differences in demographics between the swing CD and the rest of its nation, we estimated a pair of statistical models. These models estimate the difference between residents of the swing CD and the rest of their nation, controlling for differences in partisan composition, educational attainment, age, and race/ethnicity.
TX-15 residents were significantly more concerned than the rest of El Norte about January 6, voter disenfranchisement and efforts by some in Congress to overturn the 2020 election. Interestingly, they were also significantly more concerned about social media censorship and more likely to support voter ID laws. MI-8 residents were also more likely to view most threats as a major threat to democracy than other residents of Yankeedom, all else equal. However, these differences were substantively smaller and many (including efforts to overturn the election, canceling history, social media censorship, and voter fraud) were not statistically significant.
Back to the original post.
